AI Data Center Expansion Strains US Power and Water Infrastructure

The rapid expansion of AI data centers is straining United States water and power infrastructure, leading to increased utility costs and heightened blackout risks. Recent reports indicate that peak water demand from these facilities can be 6 to 30 times higher than annual averages, draining aquifers in towns such as Norwalk, Iowa, and raising water bills by 33% in Newton County, Georgia. This resource strain is coupled with a surge in energy costs; wholesale power prices in the PJM Interconnection market jumped nearly 76% year-over-year in the first quarter of 2026.

Jeremiah Johnson, a lead author of a study in Environmental Research Letters, notes that national wholesale electricity costs could rise up to 29% by 2030, with Virginia potentially seeing spikes of 57%. To meet this demand, utilities may rely on natural gas and coal, potentially increasing CO2 emissions by 28%. While some analysts like Nic Carter argue AI has not yet significantly impacted total national generation, others warn of a structural shift, with the Electric Power Research Institute projecting demand could reach 95 gigawatts by 2030.

Public opposition has intensified, with 70% of Americans opposing data centers near their homes. In 2025, projects totaling $156 billion were blocked or stalled, and Maine recently passed legislation to block new builds over 20 megawatts until 2027. Despite this, hyperscalers continue to accelerate spending, with Moody's Ratings projecting capital expenditures for top firms to reach nearly $1 trillion by 2027 to support AI revenue growth.